Mercurial > hg
view mercurial/hook.py @ 30442:41a8106789ca
util: implement zstd compression engine
Now that zstd is vendored and being built (in some configurations), we
can implement a compression engine for zstd!
The zstd engine is a little different from existing engines. Because
it may not always be present, we have to defer load the module in case
importing it fails. We facilitate this via a cached property that holds
a reference to the module or None. The "available" method is
implemented to reflect reality.
The zstd engine declares its ability to handle bundles using the
"zstd" human name and the "ZS" internal name. The latter was chosen
because internal names are 2 characters (by only convention I think)
and "ZS" seems reasonable.
The engine, like others, supports specifying the compression level.
However, there are no consumers of this API that yet pass in that
argument. I have plans to change that, so stay tuned.
Since all we need to do to support bundle generation with a new
compression engine is implement and register the compression engine,
bundle generation with zstd "just works!" Tests demonstrating this
have been added.
How does performance of zstd for bundle generation compare? On the
mozilla-unified repo, `hg bundle --all -t <engine>-v2` yields the
following on my i7-6700K on Linux:
engine CPU time bundle size vs orig size throughput
none 97.0s 4,054,405,584 100.0% 41.8 MB/s
bzip2 (l=9) 393.6s 975,343,098 24.0% 10.3 MB/s
gzip (l=6) 184.0s 1,140,533,074 28.1% 22.0 MB/s
zstd (l=1) 108.2s 1,119,434,718 27.6% 37.5 MB/s
zstd (l=2) 111.3s 1,078,328,002 26.6% 36.4 MB/s
zstd (l=3) 113.7s 1,011,823,727 25.0% 35.7 MB/s
zstd (l=4) 116.0s 1,008,965,888 24.9% 35.0 MB/s
zstd (l=5) 121.0s 977,203,148 24.1% 33.5 MB/s
zstd (l=6) 131.7s 927,360,198 22.9% 30.8 MB/s
zstd (l=7) 139.0s 912,808,505 22.5% 29.2 MB/s
zstd (l=12) 198.1s 854,527,714 21.1% 20.5 MB/s
zstd (l=18) 681.6s 789,750,690 19.5% 5.9 MB/s
On compression, zstd for bundle generation delivers:
* better compression than gzip with significantly less CPU utilization
* better than bzip2 compression ratios while still being significantly
faster than gzip
* ability to aggressively tune compression level to achieve
significantly smaller bundles
That last point is important. With clone bundles, a server can
pre-generate a bundle file, upload it to a static file server, and
redirect clients to transparently download it during clone. The server
could choose to produce a zstd bundle with the highest compression
settings possible. This would take a very long time - a magnitude
longer than a typical zstd bundle generation - but the result would
be hundreds of megabytes smaller! For the clone volume we do at
Mozilla, this could translate to petabytes of bandwidth savings
per year and faster clones (due to smaller transfer size).
I don't have detailed numbers to report on decompression. However,
zstd decompression is fast: >1 GB/s output throughput on this machine,
even through the Python bindings. And it can do that regardless of the
compression level of the input. By the time you have enough data to
worry about overhead of decompression, you have plenty of other things
to worry about performance wise.
zstd is wins all around. I can't wait to implement support for it
on the wire protocol and in revlogs.
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Fri, 11 Nov 2016 01:10:07 -0800 |
parents | 5564fcd031df |
children | 39d13b8c101d |
line wrap: on
line source
# hook.py - hook support for mercurial # # Copyright 2007 Matt Mackall <mpm@selenic.com> # # This software may be used and distributed according to the terms of the # GNU General Public License version 2 or any later version. from __future__ import absolute_import import os import sys import time from .i18n import _ from . import ( demandimport, error, extensions, util, ) def _pythonhook(ui, repo, name, hname, funcname, args, throw): '''call python hook. hook is callable object, looked up as name in python module. if callable returns "true", hook fails, else passes. if hook raises exception, treated as hook failure. exception propagates if throw is "true". reason for "true" meaning "hook failed" is so that unmodified commands (e.g. mercurial.commands.update) can be run as hooks without wrappers to convert return values.''' if callable(funcname): obj = funcname funcname = obj.__module__ + "." + obj.__name__ else: d = funcname.rfind('.') if d == -1: raise error.HookLoadError( _('%s hook is invalid: "%s" not in a module') % (hname, funcname)) modname = funcname[:d] oldpaths = sys.path if util.mainfrozen(): # binary installs require sys.path manipulation modpath, modfile = os.path.split(modname) if modpath and modfile: sys.path = sys.path[:] + [modpath] modname = modfile with demandimport.deactivated(): try: obj = __import__(modname) except (ImportError, SyntaxError): e1 = sys.exc_info() try: # extensions are loaded with hgext_ prefix obj = __import__("hgext_%s" % modname) except (ImportError, SyntaxError): e2 = sys.exc_info() if ui.tracebackflag: ui.warn(_('exception from first failed import ' 'attempt:\n')) ui.traceback(e1) if ui.tracebackflag: ui.warn(_('exception from second failed import ' 'attempt:\n')) ui.traceback(e2) if not ui.tracebackflag: tracebackhint = _( 'run with --traceback for stack trace') else: tracebackhint = None raise error.HookLoadError( _('%s hook is invalid: import of "%s" failed') % (hname, modname), hint=tracebackhint) sys.path = oldpaths try: for p in funcname.split('.')[1:]: obj = getattr(obj, p) except AttributeError: raise error.HookLoadError( _('%s hook is invalid: "%s" is not defined') % (hname, funcname)) if not callable(obj): raise error.HookLoadError( _('%s hook is invalid: "%s" is not callable') % (hname, funcname)) ui.note(_("calling hook %s: %s\n") % (hname, funcname)) starttime = time.time() try: r = obj(ui=ui, repo=repo, hooktype=name, **args) except Exception as exc: if isinstance(exc, error.Abort): ui.warn(_('error: %s hook failed: %s\n') % (hname, exc.args[0])) else: ui.warn(_('error: %s hook raised an exception: ' '%s\n') % (hname, exc)) if throw: raise if not ui.tracebackflag: ui.warn(_('(run with --traceback for stack trace)\n')) ui.traceback() return True, True finally: duration = time.time() - starttime ui.log('pythonhook', 'pythonhook-%s: %s finished in %0.2f seconds\n', name, funcname, duration) if r: if throw: raise error.HookAbort(_('%s hook failed') % hname) ui.warn(_('warning: %s hook failed\n') % hname) return r, False def _exthook(ui, repo, name, cmd, args, throw): ui.note(_("running hook %s: %s\n") % (name, cmd)) starttime = time.time() env = {} # make in-memory changes visible to external process if repo is not None: tr = repo.currenttransaction() repo.dirstate.write(tr) if tr and tr.writepending(): env['HG_PENDING'] = repo.root for k, v in args.iteritems(): if callable(v): v = v() if isinstance(v, dict): # make the dictionary element order stable across Python # implementations v = ('{' + ', '.join('%r: %r' % i for i in sorted(v.iteritems())) + '}') env['HG_' + k.upper()] = v if repo: cwd = repo.root else: cwd = os.getcwd() r = ui.system(cmd, environ=env, cwd=cwd) duration = time.time() - starttime ui.log('exthook', 'exthook-%s: %s finished in %0.2f seconds\n', name, cmd, duration) if r: desc, r = util.explainexit(r) if throw: raise error.HookAbort(_('%s hook %s') % (name, desc)) ui.warn(_('warning: %s hook %s\n') % (name, desc)) return r # represent an untrusted hook command _fromuntrusted = object() def _allhooks(ui): """return a list of (hook-id, cmd) pairs sorted by priority""" hooks = _hookitems(ui) # Be careful in this section, propagating the real commands from untrusted # sources would create a security vulnerability, make sure anything altered # in that section uses "_fromuntrusted" as its command. untrustedhooks = _hookitems(ui, _untrusted=True) for name, value in untrustedhooks.items(): trustedvalue = hooks.get(name, (None, None, name, _fromuntrusted)) if value != trustedvalue: (lp, lo, lk, lv) = trustedvalue hooks[name] = (lp, lo, lk, _fromuntrusted) # (end of the security sensitive section) return [(k, v) for p, o, k, v in sorted(hooks.values())] def _hookitems(ui, _untrusted=False): """return all hooks items ready to be sorted""" hooks = {} for name, cmd in ui.configitems('hooks', untrusted=_untrusted): if not name.startswith('priority'): priority = ui.configint('hooks', 'priority.%s' % name, 0) hooks[name] = (-priority, len(hooks), name, cmd) return hooks _redirect = False def redirect(state): global _redirect _redirect = state def hook(ui, repo, name, throw=False, **args): if not ui.callhooks: return False hooks = [] for hname, cmd in _allhooks(ui): if hname.split('.')[0] == name and cmd: hooks.append((hname, cmd)) res = runhooks(ui, repo, name, hooks, throw=throw, **args) r = False for hname, cmd in hooks: r = res[hname][0] or r return r def runhooks(ui, repo, name, hooks, throw=False, **args): res = {} oldstdout = -1 try: for hname, cmd in hooks: if oldstdout == -1 and _redirect: try: stdoutno = sys.stdout.fileno() stderrno = sys.stderr.fileno() # temporarily redirect stdout to stderr, if possible if stdoutno >= 0 and stderrno >= 0: sys.stdout.flush() oldstdout = os.dup(stdoutno) os.dup2(stderrno, stdoutno) except (OSError, AttributeError): # files seem to be bogus, give up on redirecting (WSGI, etc) pass if cmd is _fromuntrusted: if throw: raise error.HookAbort( _('untrusted hook %s not executed') % name, hint = _("see 'hg help config.trusted'")) ui.warn(_('warning: untrusted hook %s not executed\n') % name) r = 1 raised = False elif callable(cmd): r, raised = _pythonhook(ui, repo, name, hname, cmd, args, throw) elif cmd.startswith('python:'): if cmd.count(':') >= 2: path, cmd = cmd[7:].rsplit(':', 1) path = util.expandpath(path) if repo: path = os.path.join(repo.root, path) try: mod = extensions.loadpath(path, 'hghook.%s' % hname) except Exception: ui.write(_("loading %s hook failed:\n") % hname) raise hookfn = getattr(mod, cmd) else: hookfn = cmd[7:].strip() r, raised = _pythonhook(ui, repo, name, hname, hookfn, args, throw) else: r = _exthook(ui, repo, hname, cmd, args, throw) raised = False res[hname] = r, raised # The stderr is fully buffered on Windows when connected to a pipe. # A forcible flush is required to make small stderr data in the # remote side available to the client immediately. sys.stderr.flush() finally: if _redirect and oldstdout >= 0: sys.stdout.flush() # write hook output to stderr fd os.dup2(oldstdout, stdoutno) os.close(oldstdout) return res